The systematic search yielded 1640 studies. A total of 14 studies—13 observational studies and 1 randomised controlled trial (RCT)—were included in this meta-analysis (see Figure S1 in Electronic Supplementary Material), with a total of 3838 patients [8,9,10,11,12,13,14,15,16,17,18,19,20,21]. Importantly, different definitions for significant CAD were used, with cut-off values for significant lesions ranging from 50% to 75% obstruction in a major coronary artery.

A total of 1806 patients (47%) underwent TAVI with preceding PCI, whereas 2032 patients (53%) with significant CAD underwent TAVI only. Two studies included patients who underwent PCI either before TAVI or concomitantly with TAVI [11, 12]. Haemodynamic parameters (e.g. fractional flow reserve) were used in 2 studies [16, 20], and 2 studies assessed the complexity of CAD by using the SYNTAX score [11, 19]. The majority of patients (86.7%) underwent transfemoral TAVI. A balloon-expandable TAVI device was implanted in 59.6% of the patients. Characteristics and event rates of included studies are presented in Table 1 and Tables S2–S4 in the Electronic Supplementary Material.

Quality assessment of individual studies revealed serious risk of bias in all observational studies (Table S5 in Electronic Supplementary Material). This was primarily due to the absence of a standardised protocol for the decision on performing PCI prior to TAVI, unclear criteria for outcome ascertainment and/or absence of adjustment for important confounders. Visual assessment of funnel plots raised some concern for publication bias for several endpoints, but this was not confirmed by the Egger test in any case (see Figure S2 in Electronic Supplementary Material). An overview of the quality assessment and certainty of evidence as assessed by the GRADE approach is presented in Tables S5–S7 in the Electronic Supplementary Material.

All-cause mortality was similar in patients undergoing TAVI only and patients undergoing TAVI and PCI at 30 days (5.9% vs 4.7%; OR: 1.27; 95% CI: 0.91–1.77; p = 0.17; I2: 0%) (Fig. 1a; [8, 9, 11,12,13,14, 16,17,18,19, 21]) and 1 year (13.6% vs 16.4%; OR: 0.91; 95% CI: 0.64–1.29; p = 0.59; I2: 45%) (Fig. 2a; [10, 11, 13, 14, 16, 18, 20, 21]). Two studies reported on all-cause mortality > 1 year, but no significant difference was found between patients with TAVI only and those with TAVI and PCI (31.5% vs 67.7%; OR: 0.68; 95% CI: 0.42–1.08; p = 0.10; I2: 49%) (Fig. 3a; [15, 20]).

Forest plots for short-term clinical outcomes, a all-cause mortality, b cardiac death, c stroke, d myocardial infarction and e major bleeding. PCI percutaneous coronary intervention

Forest plots for mid-term clinical outcomes, a all-cause mortality, b cardiac death and c myocardial infarction. PCI percutaneous coronary intervention

Forest plot for all-cause mortality > 1 year. PCI percutaneous coronary intervention

Cardiac death rates did not differ between patients with TAVI only and those with PCI before TAVI at 30 days (3.3% vs 1.5%; OR: 1.94; 95% CI: 0.36–10.56; p = 0.45; I2: 28%) (Fig. 1b; [8, 12, 16]) and 1‑year follow-up (8.1% vs 12.3%; OR: 0.77; 95% CI: 0.19–3.13; p = 0.72; I2: 84%) (Fig. 2b; [16, 20]).

The stroke incidence at 30 days was similar between patients treated with TAVI only and those undergoing TAVI with preceding PCI (1.3% vs 2.6%; OR: 0.77; 95% CI: 0.31–1.92; p = 0.57; I2: 0%) (Fig. 1c; [9, 16, 17]). One-year stroke incidence was 4.9% and 4.6% for patients undergoing TAVI only and patients TAVI with preceding PCI, respectively [10, 16].

MI rates were similar in patients treated with TAVI only and patients undergoing TAVI and PCI at 30 days (1.1% vs 2.1%; OR: 0.50; 95% CI: 0.13–1.91; p = 0.31; I2: 0%) (Fig. 1d; [8, 10, 16]), and they remained similar at 1‑year follow-up (4.3% vs 6.0%; OR: 0.74; 95% CI: 0.21–2.66; p = 0.64; I2: 18%) (Fig. 2c; [10, 16]). Neither type of MI nor subsequent treatment was reported.

Patients treated with TAVI only had a significantly lower risk of major bleeding during the first 30 days (7.4% vs 9.4%; OR: 0.66; 95% CI: 0.46–0.94; p = 0.022; I2: 0%) (Fig. 1e; [9, 11, 12, 16, 21]). Major bleeding at 1 year was assessed in 1 study, which reported an incidence of 18.1% in patients undergoing TAVI only versus 26.1% in those with PCI before TAVI (p = 0.19) [16].

Eight studies defined significant CAD as a stenosis ≥ 70% in a major epicardial coronary artery [10,11,12, 15,16,17, 19, 21]. In line with the results of the meta-analysis on all studies, sensitivity analysis showed a significantly lower risk of major bleeding in patients treated with TAVI only compared with those undergoing TAVI and PCI (OR: 0.68; 95% CI: 0.47–0.99; p = 0.043; I2: 0%) (Figure S3 in Electronic Supplementary Material). No differences between the 2 groups were found in other clinical outcomes (Figure S3 in Electronic Supplementary Material).

Four studies used a cut-off value of 50% in their definition of significant CAD [8, 9, 13, 20]. The sensitivity analysis showed no significant difference in all-cause mortality within 30 days. However, TAVI without preceding PCI resulted in a significantly lower incidence of all-cause mortality at 1 year (OR 0.45; 95% CI: 0.28–0.74; p = 0.002; I2: 0%) (Figure S4 in Electronic Supplementary Material). No data on other clinical outcomes were available in these studies.

Moreover, the results of the meta-analysis using the random-effects model persisted in the fixed-effect models.