The MUVITAVI (MUrcia and VIgo Transcatheter Aortic Valvular Implantation) project is an observational study of patients who underwent two aortic valve replacement procedures (TAVR and SAVR) consecutively recruited between February-2018 to February 2020 from two Spanish referenced hospitals: Hospital Clínico Universitario Virgen de la Arrixaca in Murcia, and Álvaro Cunqueiro in Vigo.

The inclusion criteria were age over 50 years, and to give their informed consent to enter the study, which was approved by the Ethics Committees from each Hospital and performed in accordance with the 1964 Declaration of Helsinki and their later amendments.

Patients with single and programmed procedures were included. Exclusion criteria included: emergency or urgent procedures two or more valves treated any suspicion of endocarditis or combined coronary revascularization.

Demographics and clinical data, comorbidities, and outcomes were recruited by independently observers blinded to hematologic analysis. Patients were followed-up every 6 months. Periprocedural complications, clinical follow-up, transthoracic echocardiography (TTE) and computer tomography imaging (CT) results were incorporated using an online standardized data collection.

Samples were collected 24 h before and 48 h after the replacement in vacuum tubes containing 0.109 M sodium citrate. Platelet-poor plasma (PPP) was obtained within 2 h after extraction by centrifugation (2000 g × 15 min) at 20°C and then stored at -80°C.

As reference plasma we used a pool of 100 healthy blood donors. Moreover, as controls we also included plasma of one patient with congenital FXI deficiency caused by the p.Cys416Tyr pathogenic gene variation in heterozygosis (FXI:C 37%), and plasma of another patient with congenital FXII deficiency caused by a homozygous deletion of a single nucleotide (c.919del G) causing a frameshift (p.Val307Cysfr*44) (FXII:C 3%).

Variations of FXI levels were calculated as the ratio between FXI levels after 48 h from procedure (B) and the FXI basal levels (A), and expressed in percentages, FXI (B/A) (%).

Plasma FXII, FXI and (pre)kallikrein were evaluated by SDS-PAGE and Western Blot using procedures described elsewhere (15, 16). Two antibodies targeting FXI were used: GAFXI-AP (Enzyme Research, United Kingdom) detecting the dimer under not-reducing conditions and GAFXI-HRP (Enzyme Research, United Kingdom) that also detects FXIa under reducing conditions using dithiothreitol. Densitometry analysis of Western blot results was done with ImageJ (17).

Quantification of FXIIa-C1 inhibitor and FXIa-C1 inhibitor complexes were done with specific ELISA using nanobodies provided by ISTH Cosyne programme (18, 19). Values were represented as percentages of those observed in kaolin-full activated reference plasma.

Determination of FXI levels was also done with a functional assay using the FXIa specific chromogenic substrate SC2366 after full activation of the contact pathway with kaolin.

Activation of the coagulation cascade was evaluated by determination of antithrombin activity using a functional anti-FXa assay with heparin, bovine FXa, and S-2765 chromogenic substrate (HemosIL TH, Instrumentation Laboratory, Italy) and by quantifying thrombin-antithrombin (TAT) complexes by SDS-PAGE Western blot under reducing conditions using a polyclonal antibody (A9522, Sigma-Aldrich) and by a specific ELISA [ab108907 –Thrombin-Antithrombin Complex (TAT) Human ELISA Kit, Abcam].

The exons and flanking regions of F12 gene were amplified and sequenced as described before (20).

We recruited all embolic events happened during follow-up period. A composite end point (CEP) of thrombotic events included: any peripheral embolic event, transient ischaemic attack or stroke, and clinical or subclinical PVT. Diagnosis of subclinical PVT required at least one of the three next imaging criteria: a twofold increase in mean transaortic Doppler gradient, mobile thrombus attached to the prosthesis, or the presence of hypo attenuated leaflet thickening (HALT) with reduced leaflet motion on CT (5).

Descriptive analysis of qualitative variables included percentages. Normally distributed continuous variables were presented as means ± standard deviations (SD), whereas non-normally distributed variables were presented as median and interquartile ranges (IQR). Pearson’s Chi-Squared test and Fisher’s exact test were used for comparison of proportions or ordinal variables. Kolmogorov–Smirnov and Shapiro–Wilk tests were used for testing normality of continuous variables. Student’s-T (parametric) or Mann–Whitney-U tests were used for comparison of two means and analysis of variance (ANOVA, parametric) or Kruskal–Wallis (non-parametric) tests were used for comparison of more than two means. Apart from p-values, 95% Confidence Intervals (95% CI) were also calculated.

Statistical analysis was performed with the use of Excel ^® (Microsoft), GraphPad Prism ^® (GraphPad Software), and IBM SPSS Statistics 21 ^® (IBM SPSS Software).

A total of 232 patients were included: 155 in TAVR and 77 in SAVR group. Table 1 shows the main characteristics of the patients.

Patients in the TAVR group were older and had higher mortality risk scores and comorbidity burden than patients in the SAVR group. Likewise, the incidence of atrial fibrillation and the use of direct oral anticoagulants were significantly higher in the TAVR group. Echocardiographic parameters were comparable between both groups.

In SAVR group, 39 patients (50.6%) had a “sutureless” type of rapid deployment prosthesis: 30 Perceval ^®, and 9 Edwards Intuity ^®; while 38 (49.3%) received a “sutured” type prosthesis: 22 Edwards Perimount Magna Ease ^®, 10 Crown PRT ^®, 3 Trifecta SJM ^®, 2 Edwards Inspiris ^® and 1 Carbomedics Top Hat ^®. In TAVR group, 106 patients (68.4%) had a balloon-expandable prosthesis Edwards Sapiens ^®, and 49 (31.6%) a self-expandable valve: 29 Core Valve ^®, 16 Allegra^{TM ®}, 2 Acurate ^®, and 2 Portico ^®.

After the procedure all patients were on standardized treatment with double antiplatelet therapy (aspirin and clopidogrel). Additionally, 49 (31%) patients in TAVR group and 7 (9.4%) patients in SAVR group were anticoagulated with either antivitamin K or direct oral anticoagulant.

We firstly evaluated a potential activation of the contact pathway by the replacement of a cardiac valve. Our results show that, regardless of the procedure, it did not cause detectable activation of FXII by Western Blot (Figure 2A). Consistent with this result, neither TAVR nor SAVR caused generation of detectable kallikrein in plasma by Western Blot (Figure 2B).

Quantification of FXIIa-C1 inhibitor complexes by a specific ELISA using nanobodies revealed negligible activation of FXII either in basal samples or in samples collected after the transcatheter aortic valve replacement process (Figure 3), independently of the replacement procedure (TAVR; pre = 0.069 ± 0.066% post = 0.077 ± 0.065%; or SAVR; pre = 0.049 ± 0.044% post = 0.024 ± 0.038%).

FXI activation, assessed by two methods, Western blot analysis using reducing conditions and by a FXIa-C1 inhibitor specific ELISA using nanobodies, revealed negligible FXI activation either in basal samples or in samples collected post-procedure (Figure 4), regardless of the replacement procedure [TAVR; pre = 0.03(0.01–0.14)% post = 0.10(0.01–0.67)%; or SAVR; pre = 3.49(0.01–7.04)% post = 0.27(0.02–3.27)].

Significant variations in FXII levels were observed between patients (Figure 2A), but as expected, they correlated with the functional and common polymorphism affecting the Kozak sequence of F12.

One patient, TAVR44, an 87-year-old-male patient who received TAVR, had a complete FXII deficiency, similar to that found in a patient with congenital FXII deficiency (Supplementary Figure 1A). As expected, all other liver proteins tested (antithrombin and FXI) reached normal levels, similar to that found in the congenital FXII-deficient patient (Supplementary Figure 1B). Genetic analysis of F12 revealed a new missense mutation, not found in ExAC or 1000 Genomes, that located in exon 11: c.1277 T > G; p.Val426Gly Supplementary Figure 1C). This genetic variant, which was found in heterozygous state, was a disease-causing mutation as predicted by Mutation Taster (21). Additionally, the patient was homozygous for an intronic variant that might affect the correct splicing of this gene: c.1251 –9^°C > T, a frequent gene variation (MAF: 0.6517) with benign prediction (Supplementary Figure 1C) and homozygous T/T for the functional Kozak polymorphism (22). TAVR44 had hypertension, Charlson comorbidity index of 4, STS score of 3.36, EuroScore II of 1.94, severe aortic valve stenosis with mild insufficiency, angina and dyspnea, NYHA III, and atrial fibrillation. The patient was under oral anticoagulation (Rivaroxaban 15 mg). The patient underwent transfemoral TAVR (Edwards Sapiens N°29). No complications were recorded during the procedure. During the 6-month follow-up no adverse event was observed. FXI levels in this patient were normal according to Western blot and functional assays, and no activation of FXI was observed by Western blot or ELISA either in both samples, basal and post-TAVR procedure (Supplementary Figure 1D).

Finally, no differences in antithrombin activity were detected in basal and post-procedure samples (85 and 92%, respectively) and no TAT complexes were observed in baseline or post-procedure samples (Supplementary Figure 1B).

For the entire cohort, no significant differences were observed in the levels of FXII and FXIIa-C1 inhibitor in the samples collected before and 2 days after the valvular replacement procedure (Figures 2, 3).

No patient included in this study had FXI or (pre)kallikrein deficiency (Figures 2B,C). As described, FXI levels showed an interpersonal heterogeneity (Figure 2C). Interestingly, in 60.8% of patients the replacement of the cardiac valve reduced FXI levels in the post-procedure sample compared to the levels observed in the pre-procedure sample. Only 18 patients showed relevant increases of FXI levels (> 150% of the basal levels) after the procedure (7.8%), but in 75 cases (32.3%) FXI levels in the post-procedure sample were lower than 80% of the value observed in the baseline sample (Figure 2C). Functional assays confirmed the reduction of FXI levels in post-procedure samples (Figure 5).

The type of procedure showed a significant effect in reducing FXI levels after the valve replacement. Thus, patients who underwent SAVR showed a more severe reduction in FXI levels than TAVR [86.0% IQR (67.0–108.5) vs. 94.5% IQR (78.7–110.0); p = 0.035]. The percentage of patients with significant reduction of FXI levels in the post-procedure sample (< 80%) was also higher among SAVR patients than TAVR patients (35/78: 44.9% vs. 40/154: 26.0%, respectively; p = 0.004).

The type of bioprosthesis, balloon-expandable vs. self-expandable, in TAVR group (100.5 ± 42.9 vs. 100.0 ± 35.3; p = ns), or sutureless vs. sutured (98.1 ± 44.8 vs. 85.4 ± 29.7; p: 0.06) in SAVR patients, did not affect FXI levels during the procedure.

All-cause death at 30-day follow-up was 1.3% without differences between groups: 1 patient on TAVR and 2 patients on SAVR. Nineteen patients had bleeding requiring transfusion: 13 (8.4%) TAVR, 6 (7.8%), SAVR (p = ns) and 24 patients required permanent pacemaker: 23 (14.5%) TAVR and 1 (1.3%) SAVR (p = 0.001).

Clinical follow-up was completed in all 229 patients who survived the first month. Table 1 shows clinical complications during a period of 22.1 ± 9.9 months (median 24 months).

Thromboembolic events after TAVR or SAVR were distributed as follows: 13 patients presented with systemic embolic events (10 stroke and 3 peripheral embolism), and only 6 patients had subclinical PVT. So, the CEP of thrombotic events was reached by 19 patients without differences between groups.

We analyzed in detail the potential role of FXI variations in the complications of these procedures. Patients with thrombotic complications or periprocedural mortality showed increased FXI levels in samples collected post-procedure compared to those from the pre-procedure, although results did not show statistical significance (96.2 ± 37.3 vs. 116.0 ± 43.1, p = 0.086, Supplementary Figure 2). Moreover, patients with elevated FXI levels after the procedure (> 150% of baseline values) showed a higher incidence of thrombotic events or periprocedural mortality than that found in patients with reduced FXI [4/18 (22.2%) vs. 2/75 (2.7%); p = 0.02] (Table 2). Among those who showed reductions in the FXI levels after the procedure, only 2 out of 75 (2.7%) had embolic events and no one died during the procedure.

On the other hand, 7 patients with reduced FXI after the cardiac valve replacement suffered severe bleeding during the procedure that required transfusions (9.3%) (Table 2). The incidence of bleeding requiring transfusion was similar in both groups (Table 2).

Coagulation activation was studied by quantifying two procoagulant markers: antithrombin levels by a functional method, and TAT by ELISA. As shown in Figure 6, independently of the procedure, transcatheter valve replacement caused minor, if any, activation of the coagulation cascade.

The influence of anticoagulation on the variation of FXI level was also evaluated. 42/232 patients (18%) were anticoagulated with antivitamin K before the procedure. The ratio pre/post of FXI levels was not associated with anticoagulation (anticoagulated: 104 ± 44 vs. not anticoagulated: 95 ± 36, p = 0.113).

Finally, we have evaluated the impact on FXI of possible confounder factors such as renal and hepatic function, left ventricular ejection fraction, pulmonary hypertension and other intraoperative and post-operative complications (pericardial drainage, permanent pacemaker, severe hemorrhage, cardiac tamponade, atrial fibrillation, infections, pleural effusion), and we found no significant associations (data not shown).