Postprocedural myocardial injury and outcomes following transcatheter aortic valve implantation

Abstract

Background

 Transcatheter aortic valve implantation (TAVI) has transformed aortic stenosis treatment, yet some patients still experience complications such as post-procedural myocardial injury (PPMI). However, the prognostic significance of PPMI remains unclear. Therefore, this study aimed to investigated the association between post-TAVI cardiac troponin T (cTnT) levels and all-cause mortality.

Methods

Using Danish nationwide registries (2014–2023), we identified TAVI patients with recorded cTnT measurements before discharge. PPMI was defined as cTnT ≥ 15 times the upper limit of normal (14 ng/L). One-year mortality was analyzed using multivariable Cox regression and results were further elaborated after stratification on sex, age groups, recent PCI (within 3 months), and eGFR groups.

Results

Among 5,187 patients, 866 (16.7 %) had PPMI. Compared to those without PPMI, these patients had longer hospital stays (median 6 vs. 4 days), higher rates of pacemaker implantation (14.2 % vs. 11.4 %), and lower baseline eGFR (54 vs. 63 ml/min). At one year, the cumulative incidence of all-cause mortality was 9.0 % in patients with PPMI versus 6.6 % in those without (p < 0.01. In adjusted analysis, PPMI was associated with an increased risk of one-year mortality (HR 1.36; 95 % CI 1.04–1.77). No effect modification was found irrespective of sex, age groups, or eGFR. However, PPMI patients with recent PCI did not carry an increased rate of mortality (HR 0.88; 95 % CI 0.28–2.75).

Conclusion

PPMI following TAVI was linked to higher one-year mortality and myocardial infarction rates, highlighting the need for increased clinical awareness in this patient subgroup.

1. Introduction

Aortic stenosis is the most prevalent valvular heart disease among individuals over 65 in the Western world, most commonly treated with transcatheter aortic valve implantation (TAVI) [1]. Among various complicating factors and events, post-procedural myocardial injury (PPMI) play a significant role [2]. Given the frequent coexistence of aortic stenosis and coronary artery disease, pre-procedural revascularization has increased, with the recent Nordic Aortic Valve Intervention 3 Trial (NOTION-3) supporting its benefits [3,4]. However, the incidence of PPMI during admission varies widely depending on the definition used, [5,6] and its prognostic implications remain inconsistent [[7], [8], [9], [10], [11], [12], [13]]. These discrepancies may stem from differences in study populations, continuous advancements in TAVI techniques, the adoption of new-generation devices, and the declining use of the transapical approach. Additionally, widespread use of high-sensitivity troponin assays has lowered specificity in diagnosing PPMI, potentially contributing to conflicting findings.

A better understanding of adverse outcomes among individuals experiencing this complication could enhance clinical awareness and inform post-procedural risk stratification, possibly meriting future areas of medical intervention. Therefore, this study aimed to assess the association between post-TAVI cardiac troponin T (cTnT) levels and the risks of all-cause mortality, heart failure hospitalization, and myocardial infarction.

2. Methods

2.1. Data sources

Danish citizens are each assigned a unique personal registration number facilitating linkage between national administrative and clinical registries. Tax-funded healthcare is universally available to all Danes. This study utilized the Danish National Patient Registry (DNPR) for hospital admissions since 1977 and outpatient visits since 1995, employing ICD codes for diagnoses and Nordic Medico-Statistical Committee classifications for surgical procedures since 1996 [14]. Additionally, data from the Civil Registration System captures sex, vital status, migration, and birth dates, [15] while the Danish National Prescription Registry records drug prescriptions since 1995 [16]. The Registry of Laboratory Results for Research encompasses blood samples from hospitals and general practitioners [17].

2.2. Study population

The study population constituted all patients surviving admission with first-time TAVI between January 1, 2014 and December 31, 2023, holding at least one recorded cTnT blood sample between the date of TAVI and the date of discharge. Patients were grouped according to PPMI, defined according to the Valve Academic Research Consortium (VARC)‐2 as post‐TAVI cTnT > 15 times the upper reference limit (URL) (14 ng/L), i.e., 15 x 14 ng/L [5]. If cTnT was increased at baseline (defined as the last measured cTnT within 30 days before TAVI), a further increase in at least 50 % post-procedure was required and the peak value had to exceed the previously stated limit of 15 x 14 ng/L. If multiple cTnT samples were registered post-TAVI, only the highest value was assessed.

2.3. Follow-up and outcomes

Patients were followed from the date of TAVI discharge for three months and one year. The primary outcome was all-cause mortality. The secondary outcomes were i) admission with HF and ii) admission with myocardial infarction. The overall validity of cardiovascular diagnoses in the DNPR is high and the diagnoses of HF and myocardial infarction have been validated with positive predictive values of 76–81 % and 88–97 %, respectively [18,19].

2.4. Study variables

Patients’ medical history was determined using diagnosis codes recorded prior to the admission date, except for hypertension and diabetes, which were identified based on previously claimed drug prescriptions (Supplementary Table 1 for ICD diagnosis and procedure codes) [20,21]. Pharmacotherapy was assessed through drug prescriptions claimed within six months before admission (Supplementary Table 2 for Anatomical Therapeutic Chemical Classification System codes). Estimated glomerular filtration rate (eGFR) was calculated using the CKD-EPI (2021) equation, as recommended by the American Society of Nephrology and the National Kidney Foundation. eGFR values were reported in mL/min/1.73 m² and categorized as follows: >60 mL/min, 30–59 mL/min, and < 30 mL/min [22,23].

2.5. Statistics

Categorial variables were reported as frequencies with percentages and differences were calculated with chi square test. Continuous variables were reported with medians and IQRs with differences calculated with the Kruskall-Wallis-test. Absolute rates of all-cause mortality were examined with the reverse Kaplan-Meier function and differences were examined with the Log-Rank test. Absolute rates of first-time HF admissions and admission with myocardial infarction were evaluated with the Aalen Johanson estimator accounting for death as a competing risk and differences were calculated with Gray’s test. Patient groups were compared with multivariable Cox proportional hazard analysis and results were reported with hazard ratios (HRs) and 95 % confidence intervals (CI). The model included the following variables: sex, age, calendar year, PCI within 3 months before TAVI, known atrial fibrillation, pacemaker implantation during TAVI admission, eGFR category, ventricular arrythmia during admission, known diabetes mellitus, known heart failure, prior myocardial infarction, prior malignancy, known chronic obstructive pulmonary disease, and known liver disease. The group without PPMI served as reference in all analyses. To account for the competing risk of death, patients who died during follow-up were censored at the time of death and no longer contributed person-time at risk. The assumption of proportional hazards were investigated and fulfilled. All statistical analyses were performed using the SAS statistical software (version 9.4,Cary, NC, USA) and R (version 3.6.1 The R Foundation, Vienna,Austria). Level of statistical significance was recognized by a p-value < 0.05.

2.6. Supplementary analyses

Eight supplementary analyses was performed. First, to elucidate potential selection bias, we examined patient characteristics, in-hospital mortality, and one-year mortality. Second, outcomes were assessed in a population where patients undergoing PPM implantation during admission were excluded. Third, outcomes were examined according to the Valve Academic Research Consortium 3 (VARC-3) definition of PPMI (i.e., cTnT > 70 times the URL) [6]. Fourth, we examined the outcomes according to cTnT quartiles of the whole study cohort (≤79 ng/L (quartile 1), 80–121 ng/L (quartile 2), 122–198 ng/L (quartile 3), and ≥ 199 ng/L (quartile 4)). Fifth, we examined the outcomes in prespecified subgroups including sex, age groups (<75 years, 75–85 years, and > 85 years), PCI within 3 month before TAVI, and eGFR groups (>60 mL/min, 30–59 mL/min, and < 30 mL/min). Sixth, as outcome rates increased most drastically in the initial part of follow-up after discharge, we conducted a landmark analyses examining the event rates from 3 months after the discharge date. In these analyses, patients who died or experienced an event within the 3 months after discharge were excluded. Seventh, we examined the outcomes after excluding those with known heart failure. Eigth, we examined the outcomes after excluding those with prior myocardial infarction.

2.7. Ethics

In Denmark registry-based studies that are conducted for the sole purpose of statistics and scientific research do not require ethical approval or informed consent by law. However, the study is approved by the data responsible institute (Capital Region of Denmark – Approval number: P-2019–191) in accordance with the General Data Protection Regulation (GDPR).

3. Results

3.1. Baseline characteristics

In total, 8,117 patients underwent TAVI from 2014 to 2023. Of these, 123 (1.5 %) were excluded due to death during admission, and 2,807 (35.1 %) were excluded due to lacking cTnT measurements. Of the remaining 5,187 patients surviving TAVI admission, we identified 866 (16.7 %) with PPMI (55 % male, median age 81 years) and 4,321 (83.3 %) without PPMI (58 % male, median age 81 years). Fig. 1 illustrates the distribution of the highest measured postprocedural TnT values. Compared to patients without PPMI, patients with PPMI were admitted longer (median 6 days vs. 4 days) and more often underwent pacemaker implantation during admission (14.2 % vs. 11.4 %). An equal proportion of patients had heart failure (26.2 % vs. 27.6 %) and had undergone PCI within 3 months before the TAVI procedure (6.7 % vs. 7.4 %). Among those with an available plasma creatinine level (n = 3,164, 61 %), PPMI patients displayed a lower median baseline eGFR (54 mL/min vs. 63 mL/min) (Table 1).

Fig. 1.

Histogram of postprocedural TnT values This histogram shows an overview of the highest measured postprocedural TnT values. The dashed red line marks the value at which patients were categorized as having PPMI, i.e., 15 times the upper reference limit (14 ng/L), i.e., 15 x 14 ng/L. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1.

Baseline characteristics according to the VARC-2 definition of PPMI.

	No PPMI N = 4321	PPMI N = 866	P value
Demographics
Male sex, N (%)	2520 (58.3)	479 (55.3)	0.10
Age, median years (p25–p75)	81 (77–85)	81 (77–85)	0.19
Period			<0.01
2014–2016	646 (15.0)	248 (28.6)	−
2017–2019	1535 (35.5)	287 (33.1)	−
2020–2023	2140 (49.5)	331 (38.2)	−
cTnT, median ng/l (p25–p75)	105 (73–152)	331 (252–520)	<0.01
During TAVI admission
Pacemaker implantation, N (%)	494 (11.4)	123 (14.2)	0.02
Admission time, median days (p25-p75)	4 (2–7)	6 (4–10)	<0.01
Atrial fibrillation	32 (0.7)	7 (0.8)	0.55
Ventricular arrythmia	10 (0.2)	0 (−)	0.18
PCI within three months before admission	318 (7.4)	58 (6.7)	0.49
Medical history at any time before admission, N (%)
PCI at any time before admission	1018 (23.6)	186 (21.5)	0.90
CABG	319 (7.4)	65 (7.5)	0.90
Permanent pacemaker	448 (10.4)	87 (10.1)	0.78
Chronic heart failure	1194 (27.6)	227 (26.2)	0.39
eGFR category, ml/min			<0.01
Unknown	1692 (39.2)	331 (38.2)	−
> 60	1438 (33.3)	198 (22.9)	−
30–60	1054 (24.4)	264 (30.5)	−
< 30	137 (3.2)	73 (8.4)	−
Myocardial infarction	653 (15.1)	122 (14.1)	0.44
Atrial fibrillation	1503 (34.8)	281 (32.5)	0.19
Stroke	568 (13.2)	109 (12.6)	0.65
Peripheral vascular disease	257 (6.0)	78 (9.0)	<0.01
Diabetes	901 (20.9)	146 (16.9)	0.01
COPD	609 (14.1)	142 (16.4)	0.08
Liver disease	152 (3.5)	37 (4.3)	0.30
Malignancy	1014 (23.5)	192 (22.2)	0.41
Alcohol abuse	171 (4.0)	33 (3.8)	0.85
Pharmacotherapy, N (%)
Beta blockers	2007 (46.5)	386 (44.6)	0.31
Calcium channel blockers	1398 (32.4)	308 (35.6)	0.06
RAS inhibitors	2383 (55.2)	461 (53.2)	0.30
Loop diuretics	2083 (48.2)	433 (50.0)	0.34
Thiazids	657 (15.2)	146 (16.9)	0.21
MRA	383 (8.9)	72 (8.3)	0.60
Statins	2787 (64.5)	571 (65.9)	0.42
Aspirin	1643 (38.0)	332 (38.3)	0.86
P2Y12 inhibitors	1024 (23.7)	207 (23.9)	0.90
Oral anticoagulants	1524 (35.3)	274 (31.6)	0.04

Abbreviations: PPMI: post-procedural myocardial injury; cTnT: cardiac troponin T; TAVI: transcatheter aortic valve implantation; PCI; percutaneous coronary intervention; CABG: coronary artery bypass graft; eGFR: estimated glomerular filtration rate; COPD: chronic obstructive pulmonary disease; RAS: renin angiotensin system; MRA: mineralocorticoid receptor antagonist.

3.2. Mortality

At three months, the cumulative incidence of all-cause mortality was 2.9 % among patients with PPMI and 1.9 % among those without (p = 0.06; Fig. 2A). Adjusted analysis showed an increased three-month mortality rate in patients with PPMI compared to those without (HR 1.61; 95 % CI, 1.01–2.54; Fig. 3A).

Fig. 2.

Cumulative incidence of all-cause mortality This figure shows the absolute risk of mortality after discharge. Fig. 1A illustrates the risk within 3 months, Fig. 1b illustrates the risk within one year. The P-value for difference between groups is calculated with the log-rank test.

Fig. 3.

Adjusted rates of outcomes A. This forest plot illustrates the absolute- and adjusted rates of outcomes with three months of follow-up. B. This forest plot illustrates the absolute- and adjusted rates of outcomes with one year of follow-up.

At one year, the cumulative incidence of all-cause mortality was 9.0 % in patients with PPMI versus 6.6 % in those without (p < 0.01; Fig. 2B). This corresponded to a significantly higher adjusted one-year mortality rate in the PPMI group (HR 1.36; 95 % CI, 1.04–1.77; Fig. 3B).

3.3. Secondary outcomes

The three-month cumulative incidence of HF admissions was 3.7 % in patients with PPMI and 3.4 % in patients without PPMI (p = 0.69; Fig. 4A), with no difference in adjusted risk (HR 1.02; 95 % CI, 0.69–1.50; Fig. 3A). When extending follow-up to one year, the cumulative incidence of HF admissions was 8.4 % in patients with PPMI and 7.2 % in patients with PPMI (p = 0.20; Fig. 4B), with no difference in adjusted risk (HR 1.12; 95 % CI, 0.86–1.46; Fig. 3B).

Fig. 4.

Cumulative incidence of admission for heart failure This figure shows the absolute risk of admission for heart failure after discharge. Fig. 4A illustrates the risk within 3 months, Fig. 4b illustrates the risk within one year. The P-value for difference between groups is calculated with the log-rank test.

The three-month cumulative incidence of myocardial infarction was 0.9 % in patients with PPMI and 0.2 % in patients without PPMI (p = 0.001; Fig. 5A), corresponding to an increased adjusted risk (HR 4.59; 95 % CI, 1.76–11.99; Fig. 3A). When extending follow-up to one year, the cumulative one-year incidence of myocardial infarction was 2.2 % for patients with PPMI and 0.9 % for patients without PPMI (p = 0.001; Fig. 5B), corresponding to an increased adjusted one-year risk (HR 2.66; 95 % CI, 1.49–4.77; Fig. 3B). During the same period, 5 patients (0.6 %) with PPMI and 29 patients without PPMI underwent first-time PCI.

Fig. 5.

One-year cumulative incidence of admission for myocardial infarction This figure shows the absolute risk of admission for myocardial infarction after discharge. Fig. 5A illustrates the risk within 3 months, Fig. 5b illustrates the risk within one year. The P-value for difference between groups is calculated with the log-rank test.

3.4. Supplementary analyses

Eight supplementary analyses were performed. First, when examining those who were excluded due to lacking cTnT measurements, overall baseline characteristics were similar, with a higher proportion of patients in this group receiving TAVI in the last study period (2020–2023). Supplementary Table 3 shows the baseline characteristics of these excluded patients. In this population, the in-hospital mortality was 2.5 %. Of the 2,807 patients surviving admission, the one-year all-cause mortality was 6.4 %.

Second, when excluding those undergoing PPM implantation during admission we identified similar rates of outcomes compared with the main results (Supplementary Table 4). Third, when defining PPMI according to the VARC-3 criteria (i.e., cTnT > 70 times the URL), [6] the association between PPMI and one-year outcomes was strengthened, although the confidence intervals were wider, resulting in greater statistical uncertainty (Supplementary Table 5). Fourth, when examining the one-year outcomes according to cTnT quartiles, we identified a stepwise increased rate of outcomes across increasing quartiles. Further, in contrast to the main results, cTnT quartile 2–4 carried increased associated rates of heart failure compared with quartile 1 (Supplementary figure 6). Fifth, when stratifying the population by sex, age groups, recent PCI (within 3 months before TAVI), and eGFR categories, no statistically significant interaction was found (p for interaction ≥ 0.05). Sixth, when conducting 3-month landmark analyses, patients with PPMI still caried increased associated rates of one-year outcomes as in the main results (Supplementary Table 8). Seventh, when examining the outcomes after excluding those with known heart failure PPMI patients still carried increased one-year rates of mortality and myocaridal infarction, as shown in the main analysis (Supplementary Table 9). Eigth, when examining the outcomes after excluding those with prior myocardial infarction, PPMI still carried increased rates of one-year outcomes (Supplementary Table 10).

4. Discussion

This nationwide cohort study found that approximately 17 % of patients experienced a PPMI during TAVI admission. The main finding was that this subgroup of patients had increased associated risks of both all-cause mortality and myocardial infarction at three months and one year. These associations were consistent across different definitions of PPMI, persisted when stratifying troponin levels into quartiles, and remained robust across subgroups defined by sex, age, and kidney function.

TAVI has revolutionized the treatment of patients with aortic stenosis across all surgical risk groups. However, during the TAVI procedure, complications such as myocardial injury may occur, affecting the clinical course and post-procedural risk of adverse events. Reported rates of PPMI vary widely, ranging from 39 % to 77 % [7,8,10,11,13,24]. Some have linked PPMI to increased mortality, [9,10] while others found no significant difference between PPMI and non-PPMI patients [8,[11], [12], [13]]. Additionally, there is an absence of data on the correlation between elevated post-procedural troponin levels and nonfatal endpoints like heart failure and myocardial infarction underscoring the necessity for further investigation. Such knowledge could help quantifying the magnitude of the problem and inform post-procedural risk stratification, possibly meriting future areas of medical intervention.

In our study, TAVI was complicated by PPMI in 17 % of patients, a proportion lower than previously reported. However, 35 % of patients undergoing TAVI did not have a recorded TnT value and were therefore excluded. Consequently, the true real-world proportion of patients with PPMI is likely even lower. Supplementary analyses also demonstrated that the incidence of PPMI has declined over calendar time. The discrepancy between the proportion of PPMI observed in this study and in prior studies may be explained by the long study period, the nationwide design, and differences in cTnT recording practices across interventional centers. Overall, the study groups appeared similar in comorbidity profiles at admission and were comparable to patients excluded due to missing cTnT measurements.

Following discharge, PPMI was associated with increased risk of all-cause mortality, particularly within the first 3 months following intervention, carrying a 1.6-fold increased hazard of fatality. The elevated mortality risk among PPMI patients paralleled the worsening outcomes observed with increasing cTnT quartiles, suggesting that even modest cTnT elevations warrant careful clinical attention, including those below the formal PPMI threshold. Further, the link between PPMI and mortality remained consistent in land-mark analyses and across sex, age, and kidney function.

Our study also reported the one-year admission rates for HF and myocardial infarction. Although an increased rate of MI was observed, the clinical relevance should be interpreted in light of the very low absolute risk—approximately 1 %. While PPMI could not be associated with increased rates of HF, the impact of heightened cTnT levels was indeed significant when categorizing the population according to cTnT quartiles. Moreover, a more than two-fold increased rate of myocardial infarction was observed in the main PPMI-group. While well established in other patient cohorts, [25] the impact of cTnT on these non-fatal outcomes remain understudied in comparable TAVI populations. To the authors’ knowledge, only one existing study has addressed the relation between cTnT increase and subsequent HF events in the context of TAVI [26]. In that study, elevated cTnT post-TAVI was associated with increased risk of a composite of HF admissions and death. This association was driven by a total of 16 events over two years and when examining the outcomes separately there was not enough power to identify an association [26]. Therefore, our findings contribute novel insights into the significance of elevated cTnT levels following TAVI.

With a nationwide cohort of approximately 5,100 TAVI patients, our study presents the largest dataset to date on post-procedural cTnT, adding valuable knowledge to a relatively common complication during TAVI. While merely hypotheses generating, the results demonstrated here are indicative of PPMI’s prognostic relevance, highlighting a possible area of medical intervention in order to further enhance TAVI outcomes.

5. Limitations

Our results should be interpreted in light of several limitations. First, as with all observational studies, causality cannot be inferred, and despite careful multivariable adjustments and supplementary analyses, residual confounding remains a possibility. Second, key clinical variables such as the need for rapid pacing, procedural duration, pre-procedural coronary angiography, and type of myocardial infarction (type 1 or type 2) were not available in our dataset. Furthermore, exposure was based on troponin measurements obtained on clinical indication, which may introduce ascertainment bias. These measurements were extracted without consideration of prior cTnT values and may reflect chronic elevation in some patients. Other biomarkers such has CKMB and NT-proBNP were only recorded in a subset of the population and therefore could not be used to guide PPMI definition. Additionally, we lacked detailed information on pre-procedural coronary artery status, which could influence both troponin release and the risk of PPMI and future events such as myocardial infarction or revascularization. Third, approximately 35 % of patients were excluded due to missing TnT values. Consequently, the observed 17 % rate of PPMI after TAVI is likely artificially high and would probably be lower in a completely unselected cohort. Fourth, the cohort precedes the North European Outcomes in TAVI In Older Patients (NOTION) 3 trial, which may lead to increased rates of pre-TAVI PCI in future practice and potentially alter the incidence of PPMI—an aspect that merits further investigation. We also did not have access to echocardiographic data prior to TAVI; variables such as ejection fraction and left ventricular hypertrophy are likely to influence troponin kinetics and may modify the risk of subsequent heart failure hospitalization. Finally, electrocardiographic data and symptom status in relation to PPMI were unavailable, limiting our ability to fully characterize the clinical significance of myocardial injury in this context.

6. Conclusions

In this Danish nationwide study, TAVI was complicated by PPMI in 17 % of patients. These patients were at increased risk mortality and myocardial infarction within the first year following intervention. While more granular data on patients with myocardial damage is necessary to render more robust evidence, these findings highlight the potential for possible future medical interventions.

CRediT authorship contribution statement

Jeppe K. Petersen: Writing – original draft, Software, Methodology, Investigation, Formal analysis, Conceptualization. Lauge Østergaard: Writing – review & editing, Supervision, Methodology, Investigation. Jarl Emanuel Strange: Writing – review & editing, Methodology, Investigation. Louise Marqvard Sørensen: Writing – review & editing, Validation, Software. Ole de Backer: Writing – review & editing, Methodology, Investigation. Lars Køber: Writing – review & editing, Supervision, Methodology. Emil Fosbøl: Writing – original draft, Supervision, Methodology, Investigation, Conceptualization.

Funding

None.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Jeppe Petersen: None.

Lauge Østergaard: independent research grant from Novo Nordisk Foundation.

Jarl Emanuel Strange: None.

Louise Marqvard Sørensen: None.

Ole de Backer: Received institutional research grants and consulting fees from Abbott, Boston Scientific and Medtronic.

Lars Køber: Speaker’s honorarium from Bayer, Novartis, AstraZeneca, Boehringer, and Novo Nordisk.

Emil Fosbøl: independent research grant from Novo Nordisk Foundation.

Appendix A. Supplementary material

The following are the Supplementary data to this article: