We retrospectively included all patients with severe AS (indexed aortic valve area ≤0.6 cm²/m² or transvalvular mean gradient >40 mm Hg or peak velocity >4 m/s) treated with trans-catheter aortic valve implantation (TAVI) in the period between 2012 till 2019 at Seoul National University Hospital. Patients were divided into 2 groups according to the development of SIRS. All medical records were systematically reviewed and retrieved for patient medical history, vital status, diagnostic tests including lab values, echocardiographic and CT angiographic data as well as procedural details. Patients' clinical data was reported up to the latest follow-up from the hospital electronic medical records. Patients were excluded if they were converted to open heart surgery as the surgical trauma and extracorporeal circulation exacerbate the inflammatory response (8–10), or if they died within <48 h after TAVI, which precludes the feasibility of SIRS data assessment. Part of the study population was also included in the K-TAVI registry previously published (11). The study complied with the declaration of Helsinki and was approved by the local ethics committee.

SIRS was defined, according to the existing ACCP/SCCM consensus guidelines, as fulfilling at least two of the following four criteria within the first 48 h after TAVI as assessed by electronic monitoring records in the intensive care unit (ICU): temperature <36.0 or >38.0°C, heart rate >90 beats/min, hyperventilation as respiratory rate >20 breaths/min or P_aCO₂ <32 mmHg, leucocytic count >12 or <4 (10⁹/L) (12, 13) (Supplementary Table 1). Clinical outcomes were assessed in accordance with the standardized end-point definitions of the Valve Academic Research Consortium 2 (VARC-2) (14).

Trans-thoracic echocardiographic studies were performed for all participants at baseline before TAVI, then follow-up was performed at 1-week, 1-month, and 1-year post-TAVI. Follow-up was available in 78 (97.5%), 143 (99.3%) patients at 1-week, 66 (82.5%), 132 (92%) at 1 month, and 47 (59%), 117 (81%) at 1 year, among the SIRS and no SIRS groups, respectively. Measurements included AV study (aortic valve area (AVA), and mean pressure gradient), left ventricular ejection fraction (LVEF) determined by 2-D biplane Simpson's method, left ventricular mass index (LVMI), left atrial volume index (LAVI), tissue doppler assessment of mitral annulus for estimation of S', E', and E/E' ratio, as well as 2-D LV global longitudinal strain (2-D LV-GLS) by speckle tracking echocardiography. All measurements were standardized according to latest American Society of Echocardiography guidelines (15), and echocardiographic-related endpoints complied to definitions as per VARC-2 consensus document (14).

CT analysis was performed routinely before the TAVI procedure for proper planning of valve size, procedure, and vascular access. Post-TAVI CT was performed routinely at 1-year follow-up, according to rigorous institutional protocols, for the assessment of hypoattenuation leaflet thickening (HALT), that often indicate leaflet thrombosis formation (16), as well as measurements of different aortic dimensions. The protocol adopted was standardized throughout the study period and was largely consistent with the latest consensus document by the Society of Cardiovascular Computed Tomography (SCCT) (17). Follow-up CT studies at 1 year were available for 45 (56.3%), and 109 (75.7%) patients among the SIRS and no SIRS groups, respectively.

Continuous variables were expressed as mean ± standard deviation or median (interquartile range), while categorical variables were presented as frequencies and percentages. Continuous variables were compared by Student's t-test or non-parametric tests according to whether normally distributed or not, respectively. Chi-square or Fisher's exact tests were used to compare categorical variables. Repeated measures ANOVA was used to compare the within-groups and between-groups effect of repeated echocardiographic measurements over time and to assess group-time interaction. Logistic regression modeling was used to assess the independent predictors of SIRS among different covariates. Cox regression analysis was performed to assess the cumulative events over time among the study groups. Two models were created; one for overall mortality and the other for composite clinical efficacy VARC-2 defined endpoints. Variables with p < 0.05 on univariate analyses were included in a multivariate model. Survival analysis according to the occurrence of SIRS was determined for each of overall mortality and clinical efficacy using Kaplan-Meier (K-M) method. The log-rank test was used to express statistical difference in survival analysis. Through-out all tests, statistical significance was assumed when two-sided p < 0.05. Statistical analyses were performed using SPSS statistics version 21 (IBM Corporation, NY, USA).

A total of 224 patients were included in our analysis. Of those 80 patients (36%) developed SIRS, while 144 (64%) had no SIRS. Table 1 shows the baseline clinical characteristics of both study groups; those with SIRS vs. those with no SIRS. Both groups were comparable regarding most clinical parameters. However, those with SIRS had significantly higher STS risk score as well as higher baseline heart rate.

Table 2 shows the baseline laboratory findings of both study groups. Patients with SIRS had significantly lower baseline hemoglobin levels, while they elicited higher total leucocytic count (TLC), Neutrophil lymphocyte ratio (NLR), absolute neutrophilic count (ANC), procalcitonin, and high-sensitive C-reactive protein (HS-CRP) at baseline.

Table 2 shows post-procedural laboratory data, which were quite similar to the baseline findings, with significantly higher indices of inflammation as well as myocardial injury as evidenced by CK-MB and troponin.

There was a diversity among the valve types and generations utilized throughout the study period, which were similarly distributed between the study groups. There was no significant difference in the procedural parameters between both groups with the exception of post-implantation systolic and diastolic blood pressures which were significantly lower among the SIRS group (Table 3).

Table 4 shows the baseline and 1-year follow-up CT data for the study groups. Baseline parameters were not significantly different. The follow-up rate was significantly lower among the SIRS group (56.3 vs. 75.7%, p = 0.003). Noticeably, the incidence of HALT, on follow-up CT, was higher among the SIRS group yet not reaching statistical significance (31.1 vs. 20.2%, p = 0.1).

TTE was performed at baseline and at predetermined follow-up points (Table 5). At baseline, no significant difference was encountered among study groups except for higher PASP in the SIRS group, which faded out on follow-up. No significant difference in echocardiographic parameters was encountered upon follow-up at different time intervals. A repeated measure ANOVA was conducted to compare different echocardiographic parameters among the study groups at different time intervals (Figure 1). The within-group difference over time was significant for all echocardiographic parameters; 2D-LVEF, AV mean PG, AVA, LVMI, LAVI, and LV-GLS, while there was no significant difference between both study groups (between-groups effect) for the aforementioned parameters. Regarding the interaction of study groups over time, it was not significant except for LV-GLS which elicited significant interaction with time (p = 0.04).

A logistic regression model was created for predicting SIRS including potentially relevant laboratory and hemodynamic covariates that showed significant difference on simple comparisons and avoiding any possible multicollinearity (Table 6). On univariate analysis, the pre-procedural Hgb level, serum creatinine, TLC, pre-TAVI heart rate (HR), post-TAVI systolic and diastolic blood pressures, as well as PASP were predictors of SIRS. However, on multiple regression analysis correcting for significant covariates, only pre-TAVI TLC, pre-TAVI HR, and post-TAVI SBP were independent predictors of SIRS.

In-hospital and over more than 5-years follow-up clinical outcome data were tracked for patient groups through the hospital electronic medical records. In-hospital clinical outcomes showed that patients with SIRS had higher rates of IH death, cardiac tamponade, major, and minor bleeding episodes, as well as stage I AKI (Table 7). Median follow-up was comparable among both study groups (p = 0.4). All-cause death, cardiac death, re-hospitalization for HF, and re-hospitalization for ACS or PCI were significantly more frequent among the SIRS group (Table 7). Looking at the VARC-2 composite endpoints, early safety events (within 30 days), as well as clinical efficacy events (after 30 days), were significantly more frequent among the SIRS group (p < 0.001, p = 0.009, respectively). On the other hand, device-related events and time-related valve safety were not significantly different between both groups (p = 0.6, p = 0.2, respectively) (Table 7).

Cox-regression analyses for time-adjusted events were displayed for the outcomes of all-cause mortality, and composite endpoint of clinical efficacy (Table 8). Multiple analysis showed that for the outcome of all-cause mortality, STS score (HR = 1.1, 95% CI = 1.01–1.18, p = 0.02), Pre-TAVI Hs CRP (HR = 1.3, 95% CI = 1.05–1.51, p = 0.01), and post-TAVI peak CK-MB (HR = 1.0, 95% CI = 1.0–1.09, p = 0.03) independently predicted all-cause mortality. For the outcome of clinical efficacy, STS score was the only independent predictor (HR = 1.06, 95% CI = 1.03–1.09, p < 0.001). Of note, although SIRS showed almost three times the hazard of predicting events for the outcome of clinical efficacy, yet it only showed marginal significance (HR = 2.8, 95% CI = 0.93–8.6, p = 0.06).

Furthermore, elaborating on survival function of both study groups using K-M analysis, showed that SIRS group had more frequent time-related events for all-cause mortality as well as for the composite endpoint of clinical efficacy (log-rank p < 0.001, and =0.01, respectively) (Figure 2).

Among different parameters that were included in a prediction model for the outcome of SIRS, it was concluded that lower post-procedural BP, among other parameters, independently predicted the occurrence of SIRS. It has been previously demonstrated that systemic hypotension results in suboptimal tissue perfusion with subsequent ischemia/reperfusion injury, which in turn triggers a vicious circle of leucocyte and endothelial activation, cytokine release, and finally SIRS (20–22).

It has been postulated that general anesthesia might increase the incidence of SIRS after TAVI (19). However, general anesthesia was employed in the majority of our patient population with no significant difference between both patient groups (p = 0.4). Not to mention that the incidence of SIRS was more or less similar to that reported by Sinning et al. (7), in whom TAVI was performed without general anesthesia (conscious sedation). Nevertheless, in our study population, the duration of anesthesia was significantly longer among the SIRS group, which supports the notion that cardiac hemodynamic challenges, more frequently encountered with longer anesthesia duration, is a major determinant in the pathogenesis of SIRS.

Our study provides the largest patient cohort assessing the impact of various valve types of different generations on the incidence of SIRS. Prior studies have assessed the incidence of SIRS either utilizing only self-expanding valves (Medtronic CoreValve®) (7) or balloon-expandable valves (Edwards SAPIEN®) (19). One might expect that shear stress (23), and transient organ hypoperfusion resulting from rapid pacing with the deployment of balloon-expandable valves (19) or, on the other hand, the hemodynamic compromise with slow deployment or recapturing of self-expandable valves (7), might favor one type of valve or another. However, our data showed that there was no difference between valve types among both groups.

Previous study showed that higher leucocytic count and maximum CRP levels post-TAVI were independent predictors of declined LVEF (3). On the contrary, serial echocardiographic assessments of patients in our study, at different time intervals, did not show difference in LV function recovery between both groups, including LV subclinical dysfunction assessed by speckle-tracking derived LV-GLS. This might be explained by the overwhelming beneficial influence of the valve implantation and relief of the LV overload which supersedes the potential impact of any post-procedural inflammatory status. In fact, an early (1 week), kind of paradoxical, more rapid recovery of LV-GLS was noticed among the SIRS group, with a later catch-up at the subsequent time intervals (1 month, 1 year). A recent study, evaluating circulating inflammatory T-cell phenotypes and its association with adverse LV remodeling post-TAVI, assumed a possible role of IL-10, produced by T-cells, as an anti-inflammatory mediator, which improves cardiac function via activating fibroblasts which has a beneficial effect in the early injury phase, whereas long-term chronic activation of fibroblasts increases myocardial fibrosis resulting in adverse remodeling and functional decline (24). This hypothesis-generating assumption should be elaborated in further dedicated larger scale studies.

For the first time, we assessed the possible relation between post-TAVI SIRS and CT findings on follow-up, particularly the intriguing finding of HALT. Several hypotheses have been postulated in the pathogenesis of HALT (25–28). An inflammatory hypothesis has been implicated in the provocation of HALT (6), which in turn was suspected to trigger inflammatory process with subsequent valve leaflet degeneration (29).

We presumed that subclinical leaflet thrombosis or CT finding of HALT might be partly due to an ongoing inflammatory status that has been early manifested as SIRS. Hypothetically, a sort of chronic activation of systemic inflammatory response might contribute to this delayed finding, and SIRS might be an early trigger. Likewise, inflammatory cascades have been implicated in atherothrombotic coronary disorders and this led to several late therapeutic intervention studies aiming to abort this inflammatory milieu (30, 31). Despite the incidence of HALT in our study was more frequent in the SIRS group of patients (31 vs. 20%), yet this wasn't statistically significant. It is noteworthy that the rate of follow-up CT among SIRS group was less than that of non-SIRS group, thus underestimating the outcome.

Previous studies have elucidated increased 30-days (7, 19), and 1-year mortality (7, 18, 19) among TAVI patients eliciting ≥2 SIRS criteria. We found a higher incidence of in-hospital mortality, major and minor bleeding, and acute kidney injury among the SIRS group. Unlike our results, Schwietz et al. showed no significant difference in early mortality, although a significant increase was observed at 1-year (18). The absence of early difference in mortality in their study might be explained by the fact that they employed apical access in almost 38% of their patient population, which entails higher early in-hospital surgical risk.

According to our study, there was a higher incidence of AKI among the SIRS patients, most of them were AKI stage 1. Looking to the baseline patient data, serum creatinine was significantly higher among the SIRS group. However, baseline serum creatinine did not independently predict the occurrence of SIRS. In the study by Sinning et al. (7), SIRS was strongly associated with the development of AKI. Despite the high prevalence of CRF in their series, yet it was not associated with the occurrence of SIRS after TAVI, in concordance with our finding. Previous studies demonstrated that the inflammatory process resulting from ischemia/reperfusion injury plays a role in the pathophysiology of AKI. In TAVI patients, transient peri-procedural renal and gut hypoperfusion might result in significant cytokine release expressed as SIRS, which plays a crucial role in the pathogenesis of AKI (32, 33).

Noticeably, in our study, the incidence of major and minor bleeding was significantly higher in the SIRS group. One of the proposed contributing factors to the development of SIRS in TAVI patients was transfusion of red blood cell (RBC) units (7). This has been previously elucidated in cardiac surgery patients, and was attributed to co-administration of inflammatory mediators e.g., IL-8 which accumulates in the stored packed RBCs (34, 35). However, in our study, baseline Hgb level did not independently predict the occurrence of SIRS in a multivariable regression model. Likewise, Sinning et al. showed that, in their study cohort, the amount of RBC transfusions as well as the net drop in Hgb level didn't predict SIRS (7). This emphasizes that bleeding and subsequent blood transfusion was only a minor contributor to the occurrence of SIRS.

Over more than 5 years median follow-up, SIRS patients had significantly higher rates of overall mortality, cardiac mortality, readmissions due to HF or due to ACS/PCI. Adopting the VARC-2 composite endpoints, SIRS patients did significantly worse in terms of early safety (<30d) and late clinical efficacy (>30d), yet device success rates and time-related valve safety were not significantly different.

Our study provides the longest follow-up for the impact of SIRS on VARC-2-defined individual and composite clinical outcomes, among a real-world population with variable risk and utilizing earlier and latest generations transcatheter aortic valves. Survival analysis for all-cause mortality and VARC-2 composite endpoint of clinical efficacy showed significantly more events among the SIRS patients. This goes along with previous studies (7, 19), which provided survival analysis for 1-year only. In accordance to Sinning et al. (7), who showed a hazard ratio (95% CI) = 7.4 (3.5–15.6) among SIRS group of patients for the outcome of all-cause mortality, our SIRS group showed a hazard ratio (95% CI) = 10.5 (2.32–47.33). However, adjusted for other covariates, SIRS did not independently predict mortality. Despite SIRS showed almost 3 times the hazard of events for the composite outcome of clinical efficacy, yet it failed to independently predict this outcome as well, with only marginal significance (p = 0.06).

In summary, TAVI might frequently enhance an inflammatory reaction culminating into SIRS. This happens independent of the valve type, and despite moving toward smaller profile devices with better technological refinements, and a more minimalistic approach (conscious sedation, and more transfemoral procedures with no surgical cut-down), yet SIRS remains a devastating problem which needs all efforts to be attenuated.

Recently, there has been an uprising interest in the concept of inflammation in the pathogenesis of cardiovascular disease (36). Several late trials have tested the efficacy of specific anti-inflammatory medications in the context of atherothrombotic coronary artery disease (30, 31, 37). A recent subgroup analysis of the CANTOS (Canakinumab Anti-inflammatory Thrombosis Outcome Study) trial showed that IL-1β-targeting therapy may decrease heart failure-related hospitalization and mortality (38). Lately, a proof-of-concept study, showed, for the first time, an association of specific inflammatory phenotypes (immune signatures) with increased mortality after TAVI (39). Thus, suggesting that distinct monocyte and T-cell signatures might stand as novel biomarkers to be considered in risk stratifying patients with severe aortic stenosis undergoing TAVI, and possibly guiding potential anti-inflammatory treatment. This remains to be further elucidated in properly designed studies.